Comparison of indicators of household food insecurity using data from the 1999 national food consumption survey.

Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2006.

Information on the present situation of household food insecurity in South Africa is fragmented. There is no comprehensive study comparing different indicators of household food security. Better information on the household food security situation in South Africa would permit relevant policy formulation and better decision-making on the allocation of limited resources. The availability of a national dataset, the first South African National Food Consumption Survey data (1999) , provided the opportunity to investigate some of the issues raised above, and to contribute to knowledge on the measurement of household food security. The aim of this study was to use the data from the 1999 National Food Consumption Survey (NFCS) to : • Determine and compare the prevalence of household food insecurity using different indicators of household food security ; • Determine the overlap of households identified as food insecure by the different indicators (i.e. how many of the same households are identified as food insecure); and to • Investigate whether there was any correlation between the indicators selected . The indicators of household food security selected were: household income, household hunger experienced, and using the index child: energy and vitamin A intake (from 24 Hour Recall (24HR) and Quantified Food Frequency data), dietary diversity (from 24HR data) and anthropometric indicators stunting and underweight. The cut offs to determine food insecure household were those used in the NFCS and the cut off for dietary diversity was exploratory. The main results of the study were as follows : • The prevalence estimates of household food insecurity ranged from 10% (underweight indicator) to 70% (low income indicator). Rural areas consistently had a higher prevalence of household food insecurity than urban areas . The Free State and Northern Cape provinces had higher levels of household food insecurity, with the Western Cape and Gauteng the lower levels of household food insecurity . • Quantified Food Frequency (QFF) data yielded lower prevalence of household food insecurity estimates than 24 hour recall (24HR) data. Household food insecurity as determined by low vitamin A intakes was higher than that determined by low energy intakes for both the 24HR and QFF data . • There was little overlap with the indicators (9-52%), indicating that the same households were not being identified by the different indicators. Low dietary diversity, low income, 24HR low vitamin A intake and hunger had higher overlaps with the other indicators. Only 12 of 2826 households (0.4%) were classified by all nine indicators as food insecure. • The dataset revealed a number of statistically significant correlations. Overall , low dietary diversity, low income, 24HR low energy intake and hunger had the stronger correlations with the other indicators. Food security is a complex, multi-dimensional concept, and from the findings of this study there was clearly no single best indicator of household food insecurity status. Overall , the five better performing indicators (higher overlaps and correlations) were : low income, 24 hour recall low energy intake, 24 hour recall low vitamin A intake, low dietary diversity and hunger; this merits their use over the other selected indicators in this study. The indicator selected should be appropriate for the purpose it is being used for, e.g. estimating prevalence of food insecurity versus monitoring the long term impact of an intervention. There are other important criteria in the selection of an indicator. Income data on a national scale has the advantage of being available annually in South Africa, and this saves time and money. The 24HR vitamin A intake and 24HR energy intake indicators has as its main draw back the skill and time needed to collect and analyse the information, which increases cost and decreases sustainability. Dietary diversity and hunger have the advantage of being simple to understand, and quicker and easier to administer and analyse. It is suggested that a national food security monitoring system in South Africa uses more than one indicator, namely : 1) household income from already existing national data, 2) the potential for including a hunger questionnaire in the census should be explored, and 3) when further researched and validated, dietary diversity could also be used in national surveys.